CN116735146B - Wind tunnel experiment method and system for establishing aerodynamic model - Google Patents
Wind tunnel experiment method and system for establishing aerodynamic model Download PDFInfo
- Publication number
- CN116735146B CN116735146B CN202311008961.2A CN202311008961A CN116735146B CN 116735146 B CN116735146 B CN 116735146B CN 202311008961 A CN202311008961 A CN 202311008961A CN 116735146 B CN116735146 B CN 116735146B
- Authority
- CN
- China
- Prior art keywords
- training
- aerodynamic
- aircraft
- verification
- sample
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000002474 experimental method Methods 0.000 title claims abstract description 35
- 238000000034 method Methods 0.000 title claims abstract description 20
- 238000012549 training Methods 0.000 claims abstract description 163
- 230000005284 excitation Effects 0.000 claims abstract description 74
- 238000012795 verification Methods 0.000 claims description 75
- 230000033001 locomotion Effects 0.000 claims description 27
- 238000004364 calculation method Methods 0.000 claims description 4
- 230000009977 dual effect Effects 0.000 claims description 4
- 230000001133 acceleration Effects 0.000 claims description 3
- 238000010200 validation analysis Methods 0.000 claims description 3
- 238000010998 test method Methods 0.000 claims 1
- 238000012360 testing method Methods 0.000 abstract description 15
- 230000009286 beneficial effect Effects 0.000 abstract description 4
- 230000000694 effects Effects 0.000 abstract description 3
- 238000005259 measurement Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000010355 oscillation Effects 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- RZVHIXYEVGDQDX-UHFFFAOYSA-N 9,10-anthraquinone Chemical compound C1=CC=C2C(=O)C3=CC=CC=C3C(=O)C2=C1 RZVHIXYEVGDQDX-UHFFFAOYSA-N 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M9/00—Aerodynamic testing; Arrangements in or on wind tunnels
- G01M9/08—Aerodynamic models
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B64—AIRCRAFT; AVIATION; COSMONAUTICS
- B64F—GROUND OR AIRCRAFT-CARRIER-DECK INSTALLATIONS SPECIALLY ADAPTED FOR USE IN CONNECTION WITH AIRCRAFT; DESIGNING, MANUFACTURING, ASSEMBLING, CLEANING, MAINTAINING OR REPAIRING AIRCRAFT, NOT OTHERWISE PROVIDED FOR; HANDLING, TRANSPORTING, TESTING OR INSPECTING AIRCRAFT COMPONENTS, NOT OTHERWISE PROVIDED FOR
- B64F5/00—Designing, manufacturing, assembling, cleaning, maintaining or repairing aircraft, not otherwise provided for; Handling, transporting, testing or inspecting aircraft components, not otherwise provided for
- B64F5/60—Testing or inspecting aircraft components or systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/28—Design optimisation, verification or simulation using fluid dynamics, e.g. using Navier-Stokes equations or computational fluid dynamics [CFD]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/08—Fluids
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Fluid Mechanics (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Mathematical Physics (AREA)
- Mathematical Analysis (AREA)
- Manufacturing & Machinery (AREA)
- Transportation (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Medical Informatics (AREA)
- Mathematical Optimization (AREA)
- Algebra (AREA)
- Pure & Applied Mathematics (AREA)
- Aviation & Aerospace Engineering (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Aerodynamic Tests, Hydrodynamic Tests, Wind Tunnels, And Water Tanks (AREA)
Abstract
The application relates to the technical field of wind tunnel experiments, and provides a wind tunnel experiment method and a wind tunnel experiment system for establishing a aerodynamic model, wherein training excitation signals are applied to an aircraft; acquiring sensor data of a sensor to obtain a training sample; training the OS-ELM based on the training sample to obtain a aerodynamic model; and verifying the generalization capability of the aerodynamic model. The aerodynamic model established by the application has good prediction capability on aerodynamic force generated after excitation applied to the aircraft, and different aerodynamic models can be established aiming at different aerodynamic coefficients, so that multidirectional prediction is realized. The aerodynamic model obtained by training under the small-amplitude excitation also has a good prediction effect on aerodynamic data of the large-amplitude excitation, so that the aircraft can obtain the global aerodynamic model with good precision only by moving near the balance point, thereby being beneficial to improving the efficiency and safety of the flight test and reducing the test cost.
Description
Technical Field
The application belongs to the technical field of wind tunnel experiments, and particularly relates to a wind tunnel experiment method and system for establishing a aerodynamic model.
Background
The wind tunnel model flight experiment is a special test form, is similar to hardware in-loop simulation, and can simulate six-degree-of-freedom motion in a wind tunnel, and is mainly used for analyzing aerodynamic characteristics in the motion process of an aircraft and evaluating the performance of a control system. The conventional static and dynamic wind tunnel test is one of important modes for acquiring aerodynamic force data of an airplane, and the data measured by the test are stored in a form of a lookup table, but the aerodynamic force of the airplane cannot be accurately predicted when the airplane moves to a place beyond the lookup table due to special conditions.
Disclosure of Invention
The application aims to provide a wind tunnel experiment method and a wind tunnel experiment system for establishing a aerodynamic model, which are used for simulating the motion state of an aircraft required by the experiment by applying a training excitation signal to the aircraft to obtain a training sample, training an OS-ELM model by the training sample to obtain an online training aerodynamic model, performing generalization capability verification on the online training aerodynamic model, and obtaining a trained aerodynamic model if the online training aerodynamic model passes the verification, wherein the model has stronger nonlinear fitting and prediction capabilities and can solve the problems in the related technology. The application is realized in the following way:
a wind tunnel experiment method for establishing aerodynamic model includes:
applying a training stimulus signal to the aircraft;
acquiring motion information of an aircraft by using a sensor to obtain a training sample;
training the OS-ELM based on the training sample to obtain an online training aerodynamic model;
applying a verification stimulus signal to the aircraft; acquiring motion information of an aircraft by using a sensor to obtain a verification sample; verifying the generalization capability of the aerodynamic model based on the verification sample; if the aerodynamic model passes the verification, obtaining a trained aerodynamic model;
the verification stimulus signal is different from the training stimulus signal, and the composition parameters of the training sample and the composition parameters of the verification sample are the same.
Further, the applying a training stimulus signal to the aircraft includes:
starting a wind tunnel according to the speed pressure corresponding to the experimental set wind speed and the steady speed pressure;
adjusting the flight attitude of the aircraft to be the experimental set flight attitude;
and after the flight attitude of the aircraft is stable, transmitting a training excitation signal to the aircraft.
Further, obtaining training samples by using the sensor to obtain motion information of the aircraft, including:
continuously acquiring a plurality of sensor data of the sensor at a set first time interval;
the aerodynamic coefficient value of the aircraft is calculated using the sensor data and a training sample is formed together with parameter values affecting the aerodynamic coefficient value.
Further, the training samples comprise an initial training sample and an online training sample;
the initial training sample carries out initial training on the OS-ELM to obtain an initial training aerodynamic model;
and the online training sample carries out online training on the OS-ELM to obtain an online training aerodynamic model.
Further, the step of obtaining an initial training aerodynamic model includes:
inputting the input value of the initial training sample into the OS-ELM model to obtain the output vector of the hidden layer of the OS-ELM：
OS-ELMOutput ofVector consisting of aerodynamic coefficient values of initial training samples +.>The square difference between them is taken as the objective function:
（1），
wherein ,output weight for OS-ELM;
solving for the solution that minimizes equation (1) to obtain the initial output weight of the OS-ELM：
（2），
wherein ：M-P generalized inverse of the output of the hidden layer of the OS-ELM;
the step of obtaining the online training aerodynamic model comprises the following steps:
inputting the input value of the online training sample into an OS-ELM model to obtain the output of the OS-ELM hidden layer；
Taking the approximation error of the initial training sample and the first obtained online training sample as an objective function, wherein the method comprises the following steps:
（3），
wherein ：the pneumatic coefficient value of the online training sample is obtained for the first time;
based on the formula (3), get：
（4）；
Introducing intermediate variables：
（5），
Obtaining：
（6）；
Based on the formula (2), the formula (4) and the formula (6), the following is obtainedThe updating of (2) is as follows:
（7）；
by iteration, get the firstThe output weight is +.>：
（8）；
To the output weightAnd after convergence, obtaining the online training aerodynamic model.
Further, verifying the generalization capability of the online training aerodynamic model includes:
inputting the input value of the verification sample into an online training aerodynamic model to obtain an output aerodynamic coefficient value;
comparing the output aerodynamic coefficient value with aerodynamic coefficient values in a validation sample;
if the difference value between the output aerodynamic coefficient value and the aerodynamic coefficient value in the verification sample is within a preset range, verification is passed; if the difference between the output aerodynamic coefficient value and the aerodynamic coefficient value in the verification sample exceeds a preset range, the verification is failed.
Further, if the verification is not passed, redesigning a training sample and/or redesigning the OS-ELM, reapplying an excitation signal to the aircraft, retraining to obtain an online aerodynamic model, and carrying out generalization capability verification again.
Further, the motion information of the aircraft comprises linear acceleration of the aircraft on x, y and z axes, angular velocity around the x, y and z axes and dynamic pressure;
the aerodynamic coefficient includes at least one of a lift coefficient, a drag coefficient, a side force coefficient, a roll torque coefficient, a yaw torque coefficient, and a pitch torque coefficient.
Further, the training excitation signal and the verification excitation signal each comprise an orthogonal multi-sinusoidal excitation signal, a dual square wave, a 3211 square wave signal, a 211 square wave excitation and a sweep signal.
The application also provides a system for executing the wind tunnel experiment method for establishing the aerodynamic model, which comprises the following steps:
an aircraft for performing the experiment;
a sensor for measuring sensor data of the aircraft;
the excitation signal module is used for applying training excitation signals and verification excitation signals to the aircraft;
the acquisition module is used for acquiring sensor data of the sensor;
the calculation module is used for calculating the aerodynamic coefficient value of the aircraft based on the sensor data, forming a training sample and a verification sample according to preset conditions, completing the training of the online training aerodynamic model and verifying the generalization capability of the online training aerodynamic model.
The technical scheme adopted by the application can achieve the following beneficial effects:
the training excitation signal is applied to the aircraft, so that the motion information of the aircraft is obtained by using the sensor, a training sample is obtained, the OS-ELM is trained based on the training sample, an online training aerodynamic model is obtained, and the generalization capability verification is carried out on the online training aerodynamic model, so that the aerodynamic model after training is obtained. The OS-ELM is adopted for training, and the OS-ELM has the advantages of high training speed and high generalization capability, so that the cost of model establishment can be saved.
The method comprises the steps of performing generalization capability verification on an online training aerodynamic model by using verification excitation signals different from training excitation signals, wherein the verification excitation signals of the training excitation signals are excitation signals of different types, so that waveforms are different, the motion histories of the excited aircraft are also different, and the online training aerodynamic model is used for predicting aerodynamic forces applied to the aircraft when the aircraft performs other types of motions so as to prove the expansibility of the model.
The aerodynamic model established by the application has good prediction capability on aerodynamic force generated after excitation applied to the aircraft, and different aerodynamic models can be established aiming at different aerodynamic coefficients, so that multidirectional prediction is realized. The aerodynamic model obtained by training under the small-amplitude excitation also has a good prediction effect on aerodynamic data of the large-amplitude excitation, so that the aircraft can obtain the global aerodynamic model with good precision only by moving near the balance point, thereby being beneficial to improving the efficiency and safety of the flight test and reducing the test cost.
Drawings
In order to more clearly illustrate the embodiments of the application or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of a wind tunnel experimental method for establishing a aerodynamic model according to an embodiment of the application;
FIG. 2 is a schematic diagram of a wind tunnel experiment system for creating aerodynamic model according to an embodiment of the present application;
in the figure: 100-aircraft; 200-a sensor; 300-an excitation signal module; 310-console; 320-servo steering engine; 400-an acquisition module; 500-a computing module; 600-a safety protection module; 700-wind tunnel.
Detailed Description
Aspects of the application will be described more fully hereinafter with reference to the accompanying drawings. This application may, however, be embodied in many different forms and should not be construed as limited to any specific structure or function presented throughout this application. Rather, these aspects are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the application to those skilled in the art. Based on the teachings herein one skilled in the art will recognize that the scope of the present application is intended to cover any aspect disclosed herein, whether alone or in combination with any other aspect of the present application. For example, any number of the apparatus or implementations set forth herein may be implemented. In addition, the scope of the present application is intended to encompass other structures, functions, or devices or methods implemented using structures and functions in addition to the aspects of the application set forth herein. It should be understood that it may embody any aspect disclosed herein by one or more elements of the claims.
The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. The terms "comprises," "comprising," and/or "includes" when used herein, specify the presence of stated features, steps, operations, and/or models, but do not preclude the presence or addition of one or more other features, steps, operations, or models.
All terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art unless otherwise defined. It should be noted that the terms used herein should be construed to have meanings consistent with the context of the present specification and should not be construed in an idealized or overly formal manner.
The specific meaning of the above terms in the present application can be understood by those of ordinary skill in the art according to the specific circumstances.
In order to make the day, technical solutions and advantages of the present application more apparent, the present application will be further described in detail below with reference to the accompanying drawings and examples.
In the prior art, the conventional static and dynamic wind tunnel test is one of important modes for acquiring aerodynamic force data of an airplane, the data measured by the test are stored in a form of a lookup table, so that the control law is inconvenient to use, the data extension capability is poor, and the aerodynamic force of the airplane cannot be accurately predicted when the airplane moves to a place beyond the lookup table due to special conditions. The conventional wind tunnel test is limited by the support and the movement mechanism, so that the real movement state of the aircraft is difficult to completely simulate. If more complete pneumatic data is required, a large number of test orders are required, so that the test period is long and the test flow is complex.
The inventor finds that through applying a training excitation signal to the aircraft in a plurality of experiments, the motion information of the aircraft is obtained by using the sensor, a training sample is obtained, the OS-ELM is trained based on the training sample, an online training aerodynamic model is obtained, and generalization capability verification is carried out on the online training aerodynamic model, so that the aerodynamic model after training is completed is obtained. The OS-ELM is adopted for training, and the OS-ELM has the advantages of high training speed and high generalization capability, so that the cost of model establishment can be saved.
The method comprises the steps of performing generalization capability verification on an online training aerodynamic model by using verification excitation signals different from training excitation signals, wherein the verification excitation signals of the training excitation signals are excitation signals of different types, so that waveforms are different, the motion histories of the excited aircraft are also different, and the online training aerodynamic model is used for predicting aerodynamic forces applied to the aircraft when the aircraft performs other types of motions so as to prove the expansibility of the model.
The aerodynamic model established by the application has good prediction capability on aerodynamic force applied to the aircraft after excitation, and different aerodynamic models can be established aiming at different aerodynamic coefficients, so that multidirectional prediction is realized. The aerodynamic model obtained by training under the small-amplitude excitation also has a good prediction effect on aerodynamic data of the large-amplitude excitation, so that the aircraft can obtain the global aerodynamic model with good precision only by moving near the balance point, thereby being beneficial to improving the efficiency and safety of the flight test and reducing the test cost.
The application relates to a wind tunnel experiment method and a wind tunnel experiment system for establishing a aerodynamic model. It should be noted that: the reference numerals of the method steps of the present application are not intended to limit the order thereof, but rather to distinguish between the different steps.
Referring to fig. 1, fig. 1 is a flow chart of a wind tunnel experimental method for establishing a aerodynamic model according to an embodiment of the application.
In this embodiment, the method may include the steps of:
s100, applying a training excitation signal to the aircraft 100;
s110, starting the wind tunnel 700 according to the speed and the pressure corresponding to the experimental set wind speed and the steady speed and pressure;
before the experiment starts, the sensor 200 needs to be installed in the aircraft 100 and the wind tunnel 700, and the sensor 200 is installed at different positions according to different functions; the servo steering engine 320 is installed on the aircraft 100, then the aircraft 100 is connected with the safety protection module 600, and then the wind tunnel 700 is started according to the speed and the pressure corresponding to the experimentally set wind speed and the steady speed and pressure.
S120, adjusting the flight attitude of the aircraft 100 to the experimental set flight attitude;
when the wind tunnel is started, the aircraft 100 can swing under the action of wind force, at the moment, a control signal is applied to the aircraft 100, a rudder deflection command is calculated by a flight control law according to sensor data and the control signal fed back by the sensor 200 on the aircraft 100, and then the rudder deflection command is sent to the servo steering engine 320, so that the flight attitude of the aircraft 100 is controlled and adjusted through the servo steering engine 320, and the flight attitude required by an experiment is achieved.
S130, after the flight attitude of the aircraft 100 is stable, a training excitation signal is sent to the aircraft 100.
After the flying attitude of the aircraft 100 is stabilized, a training excitation signal is applied to the aircraft 100, and the training excitation signal is superposed on the control signal, so that the aircraft 100 is changed from a stable state to an oscillating motion state, and the training excitation signal comprises an orthogonal multi-sine excitation signal, a dual square wave, a 3211 square wave signal, a 211 square wave excitation signal, a sweep frequency signal and the like.
S200, acquiring motion information of an aircraft by using a sensor to obtain a training sample;
the motion information of the aircraft comprises linear acceleration of the aircraft on x, y and z axes, angular velocity around the x, y and z axes, dynamic pressure and the like, and the steps of obtaining training samples are as follows:
s210, continuously acquiring a plurality of sensor data of the sensor 200 on the aircraft 100 at a set first time interval;
before the experiment starts, a first time interval is determined according to the purpose of the experiment, when the aircraft 100 is in an oscillation state, a plurality of sensor data are continuously acquired according to the first time interval, the specific number of the sensor data is confirmed according to the purpose of the experiment, and the specific number can be adjusted according to the follow-up aerodynamic model verification condition.
S220, calculating the aerodynamic coefficient value of the aircraft 100 by using the sensor data, and forming a training sample together with the parameter value affecting the aerodynamic coefficient value;
selecting characteristics influencing the pneumatic coefficient, obtaining an expression of a characteristic vector by mixing the characteristics influencing the pneumatic coefficient, a higher-order item of the characteristics, a coupling item after mixing the characteristics and the like, calculating the pneumatic coefficient value based on the acquired sensor data at a first time interval, and substituting the sensor data into the expression of the characteristic vector to obtain an input value of a training sample, wherein the input value of the training sample and the calculated pneumatic coefficient value together form a verification sample.
The aerodynamic coefficient includes at least one of a lift coefficient, a drag coefficient, a side force coefficient, a roll torque coefficient, a yaw torque coefficient, and a pitch torque coefficient.
The training samples comprise initial training samples and online training samples; initial training is carried out on the OS-ELM by the initial training sample to obtain an initial training aerodynamic model; and carrying out online training on the OS-ELM by using the online training sample to obtain an online training aerodynamic model.
S300, training the OS-ELM based on a training sample to obtain an online training aerodynamic model;
first, the step of obtaining an initial training aerodynamic model comprises:
inputting the input value of the initial training sample into the OS-ELM model to obtain the output vector of the hidden layer of the OS-ELM；
Output of OS-ELMVector consisting of aerodynamic coefficient values of initial training samples +.>The square difference between them is taken as the objective function:
（1），
wherein ,output weight for OS-ELM;
solving to minimize equation (1)To obtain the initial output weight of the OS-ELM：
（2），
wherein ：the M-P generalized inverse of the output of the hidden layer of the OS-ELM.
Secondly, the step of obtaining an online training aerodynamic model comprises the following steps:
inputting the input value of the online training sample into an OS-ELM model to obtain the output of the OS-ELM hidden layer；
Taking the approximation error of the initial training sample and the first obtained online training sample as an objective function, wherein the method comprises the following steps:
（3），
wherein ：the pneumatic coefficient value of the online training sample is obtained for the first time;
based on the formula (3), get：
（4）；
Introducing intermediate variables：
（5），
Obtaining：
（6）；
Based on the formula (2), the formula (4) and the formula (6), the following is obtainedThe updating of (2) is as follows:
（7）；
by iteration, get the firstOutput weight when the online training sample arrives>：
（8）；
Weight to be outputAfter convergence, an online training aerodynamic model is obtained, and the output weight can be observed through online drawing software>And under the convergence condition, the efficiency is improved.
S400, applying a verification excitation signal to the aircraft; acquiring motion information of an aircraft by using a sensor to obtain a verification sample; verifying the generalization capability of the aerodynamic model based on the verification sample; if the aerodynamic model passes the verification, obtaining a trained aerodynamic model;
s410, applying a verification stimulus signal to the aircraft 100;
in the case where the flying attitude of the aircraft 100 is experimentally set and kept stable, a verification excitation signal is applied to the aircraft 100, the verification excitation signal being superimposed on the aforementioned control signal as with the training excitation signal to change the aircraft 100 from a steady state to an oscillating motion state, the verification excitation signal being different from the training excitation signal, the verification excitation signal being a different type of excitation signal from the aforementioned training excitation signal, the verification excitation signal including an orthogonal multi-sinusoidal excitation signal, a dual square wave, a 3211 square wave signal, a 211 square wave excitation, and a sweep signal. Because the waveforms of the excitation signals of different types are different, the motion histories of the excited aircraft are different, the aerodynamic force suffered by the aircraft 100 in other types of motion is predicted by using the established aerodynamic model, and the prediction result is compared with the calculated aerodynamic coefficient to verify the generalization capability of the aerodynamic model.
S420, acquiring motion information of the aircraft by using a sensor to obtain a verification sample;
before the experiment starts, a second time interval is determined according to the experiment purpose, and when the aircraft 100 is in the oscillation state, a plurality of sensor data are continuously acquired according to the second time interval, and the specific number of the sensor data can be confirmed according to the experiment purpose and can be adjusted according to the follow-up aerodynamic model verification condition.
The composition parameters of the verification sample and the composition parameters of the training sample are the same, the characteristics influencing the pneumatic coefficient are selected, the characteristic influencing the pneumatic coefficient, the coupling item after mixing the high-order item of the characteristic and the characteristic, and the like are used for obtaining the expression of the characteristic vector, the pneumatic coefficient value is obtained by calculation based on the acquired sensor data of the second time interval, meanwhile, the acquired sensor data of the second time interval is substituted into the expression of the characteristic vector to obtain the input value of the verification sample, and the input value of the verification sample and the calculated pneumatic coefficient value are combined into the verification sample.
The first time interval and the second time interval may be equal or unequal, which is not limited in the embodiment of the present application.
The aerodynamic coefficient includes at least one of a lift coefficient, a drag coefficient, a side force coefficient, a roll torque coefficient, a yaw torque coefficient, and a pitch torque coefficient.
S430, verifying the generalization capability of the aerodynamic model based on the verification sample;
inputting the input value of the verification sample into an online training aerodynamic model to obtain an output aerodynamic coefficient value;
comparing the output aerodynamic coefficient value with the aerodynamic coefficient value in the verification sample, and observing the curves of the output aerodynamic coefficient value and the verification aerodynamic coefficient value in real time through online drawing software, so that the verification result can be judged in time, and the efficiency is improved;
if the difference between the output aerodynamic coefficient value and the aerodynamic coefficient value in the verification sample is within a preset range, the verification is passed; if the difference between the output aerodynamic coefficient value and the aerodynamic coefficient value in the verification sample exceeds the preset range, the verification is failed.
If the verification is not passed, redesigning the training sample and/or redesigning the OS-ELM, reapplying the excitation signal to the aircraft, retraining to obtain an online aerodynamic model, and carrying out generalization capability verification again.
Reasons for the failure of the aerodynamic model to verify and ways to adjust the defect include, but are not limited to:
1. the output aerodynamic coefficient value curve has similar trend to the aerodynamic coefficient value verification curve, but is in obvious step saw-tooth shape and is not smooth. The reason may be that the nonlinear fitting capability is weak due to the fact that the number of hidden layer nodes of the OS-ELM is too small, and defects can be adjusted by increasing the number of hidden layer nodes.
2. There is a sudden change in the curve values of the output aerodynamic coefficient values, and particularly large or small errors may occur at certain points. The reason may be that the feature selection is incorrect and the features affecting the pneumatic coefficients need to be re-analyzed to adjust the training samples.
3. The trend and the numerical value of the curve for outputting the aerodynamic coefficient value cannot be matched with those of the curve for verifying the aerodynamic coefficient value. The reasons may be that:
(1) The training time is too short, the weight of the output layer is not completely converged, the number of training samples is required to be increased, and the training time is prolonged.
(2) The hidden layer nodes are too many, so that the output layer weights cannot be converged, and the hidden layer nodes need to be reduced.
4. The output aerodynamic coefficient value does not differ much from the validation aerodynamic coefficient value, but the trend is not. The reason may be that there are too many higher order terms in the feature vector, or too many hidden layer nodes, resulting in an overfitting, requiring a reduction in higher order terms in the feature vector, or a reduction in hidden layer nodes.
The embodiment of the application also provides a wind tunnel experimental system for establishing a aerodynamic model, referring to fig. 2, fig. 2 is a schematic diagram of the wind tunnel experimental system for establishing a aerodynamic model, the system comprises:
aircraft 100, mounted within wind tunnel 700, for performing an experiment;
a sensor 200 for measuring sensor data of the aircraft 100; the sensor 200 may be a sensor commonly used in wind tunnel experiments, including: an inertial measurement unit, a attitude reference system, an optical measurement sensor, etc., wherein the inertial measurement unit, the attitude reference system are disposed within the aircraft 100 and the optical measurement sensor are disposed within the wind tunnel 700.
An excitation signal module 300 for applying training excitation signals and verification excitation signals to the aircraft 100; the excitation signal module 300 may include a console 310 and a servo steering engine 320, where the console 310 may be provided with an enable switch and a plurality of excitation signal switches, one for each excitation signal type. Servo steering engine 320 is mounted within aircraft 100. When excitation is applied, the enabling switch is turned on, then an excitation signal to be applied is selected and the corresponding switch is toggled, and the servo steering engine 320 executes relevant actions according to the received instructions.
An acquisition module 400 for acquiring sensor data of the sensor 200 on the aircraft 100, obtaining a training sample and a verification sample;
the calculation module 500 is connected with the acquisition module 400, and is used for calculating the aerodynamic coefficient value of the aircraft based on the sensor data, forming a training sample and a verification sample according to preset conditions, and completing the training of the online training aerodynamic model and verifying the generalization capability of the online training aerodynamic model.
Safety protection module 600, safety protection module 600 is connected with aircraft 100, and when aircraft 100 takes place out of control, safety protection module 600 can control aircraft 100, and safety protection module 600 can include the safety rope that is connected with aircraft 100, can control aircraft 100 through tightening the safety rope.
Finally, it should be noted that: the above embodiments and features of the embodiments may be combined with each other without conflict. The above embodiments are only for illustrating the technical solution of the present application, and are not limiting; although the application has been described in detail with reference to the foregoing embodiments, it will be appreciated by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not drive the essence of the corresponding technical solutions to depart from the spirit and scope of the technical solutions of the embodiments of the present application. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to fall within the scope of the application.
Claims (6)
1. The wind tunnel experimental method for establishing the aerodynamic model is characterized by comprising the following steps of:
applying a training stimulus signal to the aircraft;
acquiring motion information of an aircraft by using a sensor to obtain a training sample;
training the OS-ELM based on the training sample to obtain an online training aerodynamic model;
applying a verification stimulus signal to the aircraft; acquiring motion information of an aircraft by using a sensor to obtain a verification sample; verifying the generalization capability of the aerodynamic model based on the verification sample; if the aerodynamic model passes the verification, obtaining a trained aerodynamic model;
the verification excitation signal is different from the training excitation signal, and the composition parameters of the training sample are the same as those of the verification sample;
the method for obtaining the training sample by using the sensor to obtain the motion information of the aircraft comprises the following steps:
continuously acquiring a plurality of sensor data of the sensor at a set first time interval;
calculating aerodynamic coefficient values of the aircraft by using the sensor data, and forming training samples together with parameter values affecting the aerodynamic coefficient values;
the training samples comprise initial training samples and online training samples;
the initial training sample carries out initial training on the OS-ELM to obtain an initial training aerodynamic model;
the online training sample carries out online training on the OS-ELM to obtain an online training aerodynamic model;
the step of obtaining an initial training aerodynamic model comprises the following steps:
inputting the input value of the initial training sample into the OS-ELM model to obtain the output of the hidden layer of the OS-ELM；
Output of OS-ELMVector consisting of aerodynamic coefficient values of initial training samples +.>The square difference between them is taken as the objective function:
（1），
wherein ,output weight for OS-ELM;
solving for the solution that minimizes equation (1) to obtain the initial output weight of the OS-ELM：
（2），
wherein ：M-P generalized inverse of the output of the hidden layer of the OS-ELM;
the step of obtaining the online training aerodynamic model comprises the following steps:
inputting the input value of the online training sample into an OS-ELM model to obtain the output of the OS-ELM hidden layer；
Taking the approximation error of the initial training sample and the first obtained online training sample as an objective function, wherein the method comprises the following steps:
（3），
wherein ：the pneumatic coefficient value of the online training sample is obtained for the first time;
based on the formula (3), get：
（4）；
Introducing intermediate variables：
（5），
Obtaining：
（6）；
Based on the formula (2), the formula (4) and the formula (6), the following is obtainedThe updating of (2) is as follows:
（7）；
by iteration, get the firstThe output weight is +.>：
（8）；
To the output weightAfter convergence, obtaining an online training aerodynamic model;
the motion information of the aircraft comprises linear acceleration of the aircraft on x, y and z axes, angular velocity around the x, y and z axes and dynamic pressure;
the aerodynamic coefficient includes at least one of a lift coefficient, a drag coefficient, a side force coefficient, a roll torque coefficient, a yaw torque coefficient, and a pitch torque coefficient.
2. A method of wind tunnel modeling aerodynamic forces according to claim 1, wherein said applying training excitation signals to an aircraft comprises:
starting a wind tunnel according to the speed pressure corresponding to the experimental set wind speed and the steady speed pressure;
adjusting the flight attitude of the aircraft to be the experimental set flight attitude;
and after the flight attitude of the aircraft is stable, transmitting a training excitation signal to the aircraft.
3. A wind tunnel test method for creating aerodynamic model according to any of claims 1-2, characterized in that verifying the generalization ability of the online training aerodynamic model comprises:
inputting the input value of the verification sample into an online training aerodynamic model to obtain an output aerodynamic coefficient value;
comparing the output aerodynamic coefficient value with aerodynamic coefficient values in a validation sample;
if the difference value between the output aerodynamic coefficient value and the aerodynamic coefficient value in the verification sample is within a preset range, verification is passed; if the difference between the output aerodynamic coefficient value and the aerodynamic coefficient value in the verification sample exceeds a preset range, the verification is failed.
4. A wind tunnel experimental method for building aerodynamic model according to claim 3, characterized in that if the verification is not passed, the training sample is redesigned and/or the OS-ELM is redesigned, the excitation signal is reapplied to the aircraft, the training is conducted again to obtain the online aerodynamic model, and the generalization capability verification is conducted again.
5. A wind tunnel experiment method for building a aerodynamic model according to claim 1, wherein the training excitation signal and the verification excitation signal comprise orthogonal multi-sinusoidal excitation signals, dual square waves, 3211 square wave signals, 211 square wave excitation and sweep signals.
6. A wind tunnel experimental system for building a aerodynamic model, characterized by being used for executing a wind tunnel experimental method for building a aerodynamic model according to any one of claims 1-5, comprising:
an aircraft for performing the experiment;
a sensor for measuring sensor data of the aircraft;
an excitation signal module for applying a training excitation signal and a verification excitation signal to the aircraft;
the acquisition module is used for acquiring sensor data of the sensor;
the calculation module is used for calculating the aerodynamic coefficient value of the aircraft based on the sensor data, forming a training sample and a verification sample according to preset conditions, completing the training of the online training aerodynamic model and verifying the generalization capability of the online training aerodynamic model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311008961.2A CN116735146B (en) | 2023-08-11 | 2023-08-11 | Wind tunnel experiment method and system for establishing aerodynamic model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311008961.2A CN116735146B (en) | 2023-08-11 | 2023-08-11 | Wind tunnel experiment method and system for establishing aerodynamic model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116735146A CN116735146A (en) | 2023-09-12 |
| CN116735146B true CN116735146B (en) | 2023-10-13 |
Family
ID=87904713
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311008961.2A Active CN116735146B (en) | 2023-08-11 | 2023-08-11 | Wind tunnel experiment method and system for establishing aerodynamic model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116735146B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110186458A (en) * | 2019-05-17 | 2019-08-30 | 天津大学 | Indoor orientation method based on OS-ELM fusion vision and Inertia information |
| CN112378619A (en) * | 2020-11-06 | 2021-02-19 | 东北财经大学 | Application of FER-FSE with ReMD-OSELM in total pressure real-time modeling in wind tunnel test stamping stage |
| CN114297779A (en) * | 2021-12-28 | 2022-04-08 | 中国航天空气动力技术研究院 | Method, system and equipment for quickly predicting aerodynamic coefficient of aircraft |
| CN114548368A (en) * | 2022-01-20 | 2022-05-27 | 广东产品质量监督检验研究院（国家质量技术监督局广州电气安全检验所、广东省试验认证研究院、华安实验室） | Modeling method, prediction model and prediction method of lithium battery temperature field prediction model based on multilayer nuclear overrun learning machine |
| CN114971032A (en) * | 2022-05-31 | 2022-08-30 | 重庆邮电大学 | Electronic nose online gas concentration prediction method based on OS-ELM |
| CN116085245A (en) * | 2023-01-31 | 2023-05-09 | 华能海南发电股份有限公司东方电厂 | Online compressor performance prediction method and system based on OS-ELM |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10030602B2 (en) * | 2014-07-22 | 2018-07-24 | The Regents Of The University Of Michigan | Adaptive machine learning method to predict and control engine combustion |
-
2023
- 2023-08-11 CN CN202311008961.2A patent/CN116735146B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110186458A (en) * | 2019-05-17 | 2019-08-30 | 天津大学 | Indoor orientation method based on OS-ELM fusion vision and Inertia information |
| CN112378619A (en) * | 2020-11-06 | 2021-02-19 | 东北财经大学 | Application of FER-FSE with ReMD-OSELM in total pressure real-time modeling in wind tunnel test stamping stage |
| CN114297779A (en) * | 2021-12-28 | 2022-04-08 | 中国航天空气动力技术研究院 | Method, system and equipment for quickly predicting aerodynamic coefficient of aircraft |
| CN114548368A (en) * | 2022-01-20 | 2022-05-27 | 广东产品质量监督检验研究院（国家质量技术监督局广州电气安全检验所、广东省试验认证研究院、华安实验室） | Modeling method, prediction model and prediction method of lithium battery temperature field prediction model based on multilayer nuclear overrun learning machine |
| CN114971032A (en) * | 2022-05-31 | 2022-08-30 | 重庆邮电大学 | Electronic nose online gas concentration prediction method based on OS-ELM |
| CN116085245A (en) * | 2023-01-31 | 2023-05-09 | 华能海南发电股份有限公司东方电厂 | Online compressor performance prediction method and system based on OS-ELM |
Non-Patent Citations (1)
| Title |
|---|
| 占健 ; 吴斌 ; 王加祥 ; 余建波 ; .基于OS-ELM的风机关键机械部件故障诊断方法.机械制造.2015,(04),第66-70页. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116735146A (en) | 2023-09-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP1141578B1 (en) | Active vibration control system identification with enhanced noise reduction | |
| Song et al. | Active model-based predictive control and experimental investigation on unmanned helicopters in full flight envelope | |
| CN111650830A (en) | Four-rotor aircraft robust tracking control method based on iterative learning | |
| KR20190068261A (en) | Design method of attitude control system for flight vehicle and computer program | |
| Marchetti et al. | Active flutter suppression analysis and wind tunnel studies of a commercial transport configuration | |
| CN114611416A (en) | LS-SVM modeling method for nonlinear unsteady aerodynamic characteristics of missile | |
| CN111427267A (en) | High-speed aircraft attack angle tracking method adopting force and moment adaptive estimation | |
| CN116735146B (en) | Wind tunnel experiment method and system for establishing aerodynamic model | |
| CN113221237B (en) | Large attack angle flutter analysis method based on reduced order modeling | |
| Suresh et al. | Direct adaptive neural flight control system for an unstable unmanned aircraft | |
| Toha et al. | Dynamic nonlinear inverse-model based control of a twin rotor system using adaptive neuro-fuzzy inference system | |
| CN115981149B (en) | Hypersonic aircraft optimal control method based on safety reinforcement learning | |
| CN116736723A (en) | Modeling and fuzzy self-adaptive sliding mode control method for elastic hypersonic aircraft under aerodynamic heat influence | |
| Haley et al. | Generalized predictive control for active flutter suppression | |
| Ivanco et al. | Generalized predictive control for active stability augmentation and vibration reduction on an aeroelastic tiltrotor model | |
| Grauer et al. | Identification of Bare-Airframe Dynamics from Closed-Loop Data Using Multisine Inputs and Frequency Responses | |
| CN110703611B (en) | Micro-gyroscope sensor terminal sliding mode control system based on recursive fuzzy neural network | |
| Drew et al. | Multi-objective gust load alleviation control designs for an aeroelastic wind tunnel demonstration wing | |
| Ricci et al. | Active control of three-surface aeroelastic model | |
| Cooper et al. | Robust adaptive disturbance compensation for ship-based rotorcraft | |
| Turner et al. | Simulation of pilot control activity for the prediction of workload ratings in helicopter/ship operations | |
| CN113189866A (en) | Method for identifying state space model of tilt rotor unmanned aerial vehicle | |
| CN115438603B (en) | Method for determining wind field dynamic response of elastic aircraft in mobile wind field environment | |
| Marretta et al. | Wing flutter suppression enhancement using a well-suited active control model | |
| CN117985236B (en) | Flight control method, device and control equipment based on fluid thrust vector |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |